Prospective isolation of human clonogenic common
myeloid progenitors
Markus G. Manz*†‡§, Toshihiro Miyamoto*‡¶, Koichi Akashi储, and Irving L. Weissman*§
*Departments of Pathology and Developmental Biology, Stanford University School of Medicine, Stanford, CA 94305; and 储Department of Cancer
Immunology and AIDS, Dana–Farber Cancer Institute, Boston, MA 02115
Contributed by Irving L. Weissman, June 27, 2002
The hierarchical development from hematopoietic stem cells
to mature cells of the hematolymphoid system involves progressive loss of self-renewal capacity, proliferation ability, and
lineage potentials. Here we show the prospective isolation
of early developmental intermediates, the human clonogenic
common myeloid progenitors and their downstream progeny,
the granulocyte兾macrophage and megakaryocyte兾erythrocyte
progenitors. All three populations reside in the lineage-negative
(linⴚ) CD34ⴙCD38ⴙ fraction of adult bone marrow as well as in
cord blood. They are distinguishable by the expression of the
IL-3R␣ chain, the receptor of an early-acting hematopoietic cytokine, and CD45RA, an isoform of a phosphotyrosine phosphatase
involved in negative regulation of cytokine signaling. Multipotent
progenitors, early lymphoid progenitors, and the here-defined
myeloid progenitors express distinct profiles of hematopoiesisaffiliated genes. The isolation of highly purified hematopoietic
intermediates provides tools to better understand developmental
programs underlying normal and leukemic hematopoiesis.
T
hroughout life, the hematolymphoid system is regenerated
from a small fraction of self-renewing hematopoietic stem
cells (HSCs). Most of the knowledge about factors that initiate
the maturation process of progenitor cells and induce restriction
to one hematopoietic lineage is achieved in principle through
two approaches: (i) the observation of natural occurring or
artificially induced genetic alterations that lead to characteristic
phenotypes, and (ii) the rigorous purification of HSC and兾or
progenitor cell populations that are tested at the clonal level for
their full developmental capacity and in short-term developmental tests for their placement in a lineage. We recently have
isolated by surface phenotype clonogenic common lymphoid
progenitors (1) and clonogenic common myeloid progenitors
(CMPs) and their myeloid downstream progeny, the granulocyte兾macrophage (GMPs) and megakaryocyte兾erythrocyte progenitors (MEPs) (2) in mouse bone marrow. These findings
together with data from other groups indicate that the commitment to either the lymphoid or myeloid lineage and downstream
to the granulocyte兾macrophage and megakaryocyte兾erythrocyte lineages are exclusive events. In humans, multiple reports
describe the enrichment of progenitor populations in bone
marrow, mobilized peripheral blood, thymus, cord blood, fetal
bone marrow, and fetal liver (3–16). However, to the best of our
knowledge, thus far only early lymphoid lineage-restricted progenitors have been isolated and evaluated rigorously for all their
developmental capacity on a clonal basis: An adult human bone
marrow lineage-negative (lin⫺) CD34⫹CD38⫹CD10⫹ cell gives
rise to B, natural killer (NK), and dendritic cells on a clonal level
and to T cells at high frequencies (7), and a cord blood
CD34⫹CD38⫺CD7⫹ cell gives rise to B, NK, and dendritic cells
(8), but both are devoid of myeloerythroid differentiation capacity. We therefore searched for distinct early myeloerythroidcommitted progenitors, the human counterparts of the described
murine CMPs, GMPs, and MEPs, in adult human bone marrow
and cord blood.
Materials and Methods
Bone Marrow and Cord Blood Samples. Fresh human bone marrow
(donor age, 20–47 years) was obtained with informed consent
11872–11877 兩 PNAS 兩 September 3, 2002 兩 vol. 99 兩 no. 18
from healthy volunteers (AllCells, Berkeley, CA), allogeneic
bone marrow donors (Department of Bone Marrow Transplantation, Stanford School of Medicine, Stanford, CA), and vertebral bodies of multiorgan cadaveric donors (Miami School of
Medicine, Miami). Cord blood was drawn with informed consent
after caesarian sections of full-term newborns (Department of
Obstetrics, Stanford School of Medicine).
Cell Preparation, Flow-Cytometric Analysis, and Cell Sorting. CD34⫹
cells were enriched from mononuclear cells by using immunomagnetic beads according to manufacturer instructions (CD34⫹selection kit, Miltenyi Biotec, Bergisch-Gladbach, Germany).
For analysis and sorting of myeloid progenitors, cells were
stained with phycoerythrin (PE)-Cy5-conjugated antibodies specific for lin markers CD3, S4.1; CD4, S3.5; CD7, CD7-6B7; CD8,
3B5; CD10, 5-1B4; CD14, TUK4; CD19, SJ25-C1 (Caltag, South
San Francisco, CA); CD2, RPA-2.10; CD11b, ICRF44; CD20,
2H7; CD56, B159; GPA, GA-R2 (Becton Dickinson–PharMingen, San Diego), and FITC-conjugated anti-CD45RA, MEM56
(Caltag), PE-conjugated anti-IL-3R␣, 9F5 (Becton Dickinson–
PharMingen), biotinylated anti-CD38, HIT2 (Caltag), and
APC-conjugated anti-CD34, HPCA-2 (Becton Dickinson–
PharMingen).
Lymphoid progenitors were stained with the same PE-Cy5conjugated lineage mixture except for the omission of anti-CD7,
anti-CD10, and the addition of anti-CD13, TUK1 (Caltag)
followed by FITC-conjugated anti-CD10 (Ancell, Bayport, MN),
PE-conjugated anti-IL-7R␣ (Immunotech, Marseilles, France),
and anti-CD38 and CD34 as described above. For additional
analysis FITC-, PE-, Texas red-, or APC-conjugated or biotinylated monoclonal antibodies against the following antigens
were used: CD45, HI30; CD13, TUK1; CD16, 3G8; CD71,
T56兾14; HLA-DR, TU36 (Caltag); CD3, HIT3a; CD4, RPA-T4,
CD8, HIT8a; CD9, M-L13; CD19, HIB19; CD32, FLI8.26;
CD33, WM53; CD41a, HIP8; CD45RA, HI100; CD56, B159;
CD64, 10.1; CD71, M-A712; CD90, A59; CD117, YB5.B8;
Glycophorin A (GPA), HIR2 (Becton Dickinson–PharMingen).
Streptavidin-conjugated Texas red or Cy7-PE (GIBCO) was
used for visualization of biotinylated antibodies. Dead cells were
excluded by propidium iodide staining. Appropriate isotypematched, irrelevant control monoclonal antibodies were used to
determine the level of background staining. Cells were sorted
and analyzed by using a highly modified dual-laser FACS
Abbreviations: HSC, hematopoietic stem cell; CMP, common myeloid progenitor; GMP,
granulocyte兾macrophage progenitor; MEP, megakaryocyte兾erythrocyte progenitor; NK,
natural killer; lin⫺, lineage-negative; PE, phycoerythrin; SCF, stem cell factor; FL, Flt3 ligand;
Epo, erythropoietin; Tpo, thrombopoietin; LTC-IC, long-term culture initiating cell; CFU,
colony-forming unit; GEMM, granulocyte兾erythroid兾macrophage兾megakaryocyte.
†Present
address: Institute for Research in Biomedicine, Via Vincenzo Vela 6, 6500 Bellinzona, Switzerland.
‡M.G.M.
and T.M. contributed equally to this work.
§To
whom reprint requests may be addressed at: Departments of Pathology and Developmental Biology, Stanford University School of Medicine, B257 Beckman Center, Stanford,
CA 94305-5428. E-mail: [email protected] or [email protected].
¶Present
address: First Department of Internal Medicine, Faculty of Medicine, Kyushu
University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.172384399
Vantage (Becton Dickinson Immunocytometry Systems, Mountain View, CA) and a dual-laser MoFlo cytometer (Cytomation,
Fort Collins, CO), both equipped with a 488-nm argon laser and
a 599-nm dye laser. All cells were double-sorted to obtain
populations that were essentially pure for the indicated surfacemarker phenotype. For single-cell and limiting-dilution assays,
the re-sort was performed by using an automatic cell-deposition
unit (ACDU) system (Becton Dickinson).
For myeloid colony formation, cells were cultured over 14 days
in Iscove’s modified Dulbecco’s medium (IMDM, GIBCO)
based methylcellulose medium (Methocult H4100, StemCell
Technologies, Vancouver) supplemented with 20% FBS (lot
no. 315853, ICN), 1% BSA, 2 mM L-glutamine, 50 ␮M 2-mercaptoethanol, antibiotics, and human IL-3 (20 ng兾ml), IL-6
(10 ng兾ml), IL-11 (10 ng兾ml), stem cell factor (SCF, 10 ng兾ml),
Flt3 ligand兾(FL, 10 ng兾ml), granulocyte兾macrophage colonystimulating factor (50 ng兾ml), thrombopoietin (Tpo, 50 ng兾ml),
and erythropoietin (Epo, 4 units兾ml) (all R & D Systems). FBS
was selected from multiple sera to obtain significant megakaryocyte development. For short-term liquid cell culture, cells were
cultured in IMDM with 10% FBS, IL-11 (10 ng兾ml), SCF (10
ng兾ml), FL (10 ng兾ml), Tpo (50 ng兾ml), and Epo (4 units兾ml).
To support the formation of B and NK cells, progenitors were
cultured for 3–5 weeks on irradiated (25 Gy) SyS-1 (Ac6) stromal
cell monolayers in IMDM supplemented with 10% FBS (HyClone), SCF (10 ng兾ml), FL (10 ng兾ml), and IL-7 (50 ng兾ml) (for
B cell readout) or IL-2 (50 ng兾ml, for NK cell readout), or IL-7
and IL-2 (for combined B and NK cell readout). Half of the
medium was replaced weekly.
Long-term culture initiating-cell (LTC-IC) assays were performed in human long-term culture medium (Myelocult H5100,
StemCell Technologies) supplemented with hydrocortisone on
irradiated M2–10B4 monolayers according to manufacturer
instructions. Cells were transferred to the above-described methylcellulose medium after 6 weeks. All cultures were incubated at
37°C in a humidified chamber under 7% CO2.
Reverse Transcription–PCR Analysis. Total RNA was purified from
1,000 double-sorted cells from each population and amplified by
reverse transcription–PCR as reported (17). Oligonucleotide
primer sequences were as reported previously and can be
provided upon request. PCR products were electrophoresed on
an ethidium bromide-stained 2.0% agarose gel. PCR analysis
was repeated at least twice for at least two separately prepared
sets of samples.
Results
Identification of Three Distinct Cell Populations with High Myeloerythroid Colony-Forming Unit (CFU) Activity in Adult Human Bone Marrow.
We first tested myeloid colony-forming potentials in CD34⫹ and
CD34⫺ human bone marrow cells and confirmed previous
findings that more than 95% of CFUs reside in the CD34⫹ cell
fraction (data not shown). Cells with multipotent, self-renewing
hematopoietic capacity are lin⫺CD34⫹CD38⫺Thy1⫹, whereas
CD34⫹CD38⫹ cells, the major fraction of all CD34⫹ cells, are
further differentiated and have lost both self-renewing capacity
and bipotent lymphomyeloid potential (18–24). The coexpression of CD10 or CD7 on CD34⫹ cells in adult bone marrow or
cord blood defines the earliest described clonal human lymphoid-committed progenitors (7, 8). To exclude multipotent
progenitors, lymphoid-restricted progenitors, and more mature
lineage low cells, we therefore searched for the earliest myeloidcommitted cells in the lin⫺ (CD2, CD3, CD4, CD7, CD8, CD10,
CD11b, CD14, CD19, CD20, CD56, and GPA) CD34⫹CD38⫹
cell fraction of adult human bone marrow.
The myeloid colony-formation efficiency in methylcellulose
Manz et al.
Fig. 1. Prospective isolation of myeloid progenitors in adult human bone
marrow. (a) The lin⫺CD34⫹CD38⫹ fraction was subdivided according to IL-3R␣
and CD45RA expression (a–c). (b) Reanalysis of the sorted populations. Percentages relative to mononuclear bone marrow cells are shown. (c) Additional
antigen expression profile on CD34⫹CD38⫺ (brown), IL-3R␣loCD45RA⫺ (red),
IL-3R␣loCD45RA⫹ (green), and IL-3R␣⫺CD45RA⫺ (blue) populations. Isotype
control gated on total lin⫺CD34⫹CD38⫹ cells (black) and marker expression on
CD34⫺ bone marrow cells (dashed line) are shown.
was 52–73% for lin ⫺ CD34 ⫹ CD38 ⫺ cells, 67–93% for
lin⫺CD34⫹CD38⫹ cells, and 0–5% for lin⫺CD34⫺CD38⫺ cells
(Figs. 1 and 2 and data not shown). We tested the expression of
multiple candidate surface antigens on the lin⫺CD34⫹CD38⫹
fraction. With antibodies against CD45RA and the IL-3R␣, the
lin⫺CD34⫹CD38⫹ fraction is clearly separable in three distinct
populations, the IL-3R␣loCD45RA⫺ (candidate CMPs), IL3R␣loCD45RA⫹ (candidate GMPs), and IL-3R␣⫺CD45RA⫺
(candidate MEPs) fraction, accounting for ⬇0.28, ⬇0.35, and
⬇0.13% of bone marrow mononuclear cells (Fig. 1 a and b). The
three populations display relatively uniform levels of the SLF
receptor c-Kit (CD117), CD13, CD33, and HLA-DR, but they
are negative for Thy-1 (CD90), FcR III (CD16), FcR II (CD32),
FcR I (CD64), CD41a, and CD9 (Fig. 1c and not shown).
The IL-3R ␣ lo CD45RA ⫺ , IL-3R ␣ lo CD45RA ⫹ , and IL3R␣⫺CD45RA⫺ fractions gave rise to distinct colony types in
methylcellulose CFU assays with the indicated cytokines (Fig. 2 a
and b). To evaluate cloning efficiency and colony types precisely,
single cells of each population from 18–27-year-old donors were
sorted in 96-well plates (Fig. 2c). HSC-enriched lin⫺CD34⫹CD38⫺
cells (288 cells, three donors), 336 lin⫺CD34⫹CD38⫹IL3R␣loCD45RA⫺ cells, 384 lin⫺CD34⫹CD38⫹IL-3R␣loCD45RA⫹,
and 384 lin⫺CD34⫹CD38⫹IL-3R␣⫺CD45RA⫺ cells (four donors
each) were plated. HSC-enriched lin⫺CD34⫹CD38⫺ cells gave
rise to all types of myeloid colonies including mixed colonies
(CFU-GEMM), burst-forming units er ythroid, CFUmegakaryocyte, CFU-megakaryocyte兾erythroid, CFU-granulocyte兾macrophage, CFU-granulocyte, and CFU-macrophage
with a total cloning efficiency of 69%. Similarly, 84% of single
IL-3R␣loCD45RA⫺ cells gave rise to a comparable distribution
of colony types. In contrast, 75% of lin⫺CD34⫹CD38⫹IL3R␣loCD45RA⫹ cells gave rise exclusively to CFU-granulocyte,
CFU-macrophage, and CFU-granulocyte兾macrophage, whereas
PNAS 兩 September 3, 2002 兩 vol. 99 兩 no. 18 兩 11873
MEDICAL SCIENCES
In Vitro Assays to Determine Differentiation Potential of Progenitors.
Fig. 2. Colony formation in methylcellulose. (a) Morphology of day-14
colonies derived from sorted progenitors (inverted light microscope, ⫻4).
(Right Insert) CFU-megakaryocyte兾erythroid (⫻10). (b) Cellular morphology
form single-picked colonies: CFU-granulocyte兾erythroid兾macrophage兾
megakaryocyte (GEMM, Left), CFU-granulocyte兾macrophage (Center), burstforming units兾erythroid (Right), and CFU-megakaryocyte (Right Insert)
(Giemsa, ⫻100 oil). (c) Clonogenic myeloid colony formation in methylcellulose. Scored were 288 wells receiving lin⫺CD34⫹CD38⫺ cells (three donors),
336 wells receiving lin⫺CD34⫹CD38⫹IL-3R␣loCD45RA⫺ cells, and 384 wells
each receiving lin⫺CD34⫹CD38⫹IL-3R␣loCD45RA⫹ and lin⫺CD34⫹CD38⫹IL3R␣⫺CD45RA⫺ cells (four donors each). CFU-MegE, CFU-megakaryocyte兾
erythroid; CFU-GM, CFU-granulocyte兾macrophage; BFU-E, burst-forming
units兾erythroid; CFU-Meg, CFU-megakaryocyte; CFU-M, CFU-macrophage;
CFU-G, CFU-granulocyte.
87% of lin⫺CD34⫹CD38⫹IL-3R␣⫺CD45RA⫺ cells produced
burst-forming units erythroid, CFU-megakaryocyte, CFUmegakaryocyte兾erythroid, and two CFU-granulocyte (0.5%). In
analogy to the defined mouse progenitors (2) we therefore
termed the IL-3R␣loCD45RA⫹ cells GMPs and the IL3R␣⫺CD45RA⫺ cells MEPs.
Morphological features of cells in each colony are shown in
Fig. 2b. In our culture condition, megakaryocytes were still
immature on day 14; a majority displayed only one or two nuclear
lobes (Fig. 2b Left), whereas scattered mature megakaryocytes
possessed lobulated polyploid nuclei (Fig. 2b Right Insert).
IL-3R␣loCD45RAⴚ Cells Give Rise to GMPs and MEPs and Contain Clonal
Progenitors of Granulocyte兾Macrophage and Megakaryocyte兾Erythrocyte Colony-Forming Cells. To test whether the oligopotent
IL-3R␣loCD45RA⫺ cells are capable to give rise to GMPs and
MEPs, 2,000–5,000 IL-3R␣loCD45RA⫺ cells were cultured on
Sys-1 stromal cells with different cytokine combinations. Devel11874 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.172384399
Fig. 3.
Lin⫺CD34⫹CD38⫹IL-3R␣loCD45RA⫺ cells are progenitors of IL3R␣loCD45RA⫹ and IL-3R␣⫺CD45RA⫺ cells and have clonal bipotent granulocyte兾macrophage and erythrocyte兾megakaryocyte potential. (a) IL3R␣loCD45RA⫺ cells give rise to both IL-3R␣loCD45RA⫹ and IL-3R␣⫺CD45RA⫺
cells after 72-h culture on Sys-1stromal cells in the presence of SCF, IL-11, FL,
Epo, and Tpo. (b) Colony readout of sorted cells from primary cultures (gates
in a) in methylcellulose. CFU-MegE, CFU-megakaryocyte兾erythroid; CFU-Meg,
CFU-megakaryocyte; BFU-E, burst-forming units erythroid; CFU-GM, CFUgranulocyte兾macrophage; CFU-M, CFU-macrophage; CFU-G, CFU-granulocyte. (c) Myeloid differentiation potential of single IL-3R␣loCD45RA⫺ cells.
Single cells were expanded in primary cultures and then transferred to methylcellulose (see Materials and Methods).
opment of both GMPs and MEPs was achieved with SCF, IL-11,
FL, Epo, and Tpo after 72 h of culture (Fig. 3a). Of note, IL-3R␣
and CD45RA expression levels were always higher in cultured
cells than in freshly isolated bone marrow cells. In four experiments with different donors, cultured cells were re-sorted on
methylcellulose (in total 250 IL-3R␣loCD45RA⫺ cells and 300
GMPs and MEPs each). As with the freshly isolated populations,
albeit with lower cloning efficiency, IL-3R␣loCD45RA⫺ cells
gave rise to all types of colonies (but CFU-GEMM), GMPs
exclusively gave rise to granulocyte兾macrophage colonies, and
MEPs gave rise to megakaryocyte兾erythrocyte colonies and four
(1.6%) granulocyte and macrophage colonies (Fig. 3b). Therefore, the IL-3R␣loCD45RA⫺ cells can give rise to functional
GMPs and MEPs and contains likely the physiologic progenitor
of both populations.
To test whether the IL-3R␣loCD45RA⫺ cell fraction contains cells that are either granulocyte兾macrophage- or erythrocyte兾megakar yocyte-committed, or whether single IL3R␣loCD45RA⫺ cells have both granulocyte兾macrophage and
megakar yocyte兾er ythrocyte potential, 300 single IL3R␣loCD45RA⫺ cells from two donors were sorted into Terasaki wells and cultured for 72 h. At this time, 242 wells
contained live cells with on average 4.8 cells per well. All wells
with ⬎1 cell (219) were transferred into secondary methylcellulose cultures (Fig. 3c). When only 2–4 secondary colonies were
formed, 15.6% of single IL-3R␣loCD45RA⫺ cells gave rise to
both granulocyte兾macrophage and megakaryocyte兾erythrocyte
Manz et al.
Fig. 4. IL-3R␣loCD45RA⫺ cells contain no lymphoid-committed progenitor
cells. (a) Development of CD19⫹ and CD56⫹ cells from 33 lin⫺ CD34⫹CD38⫹IL7R␣⫹CD10⫹ cells after 4 weeks of culture. (b) No CD19⫹ or CD56⫹ cells develop
from 5,000 IL-3R␣loCD45RA⫺ cells under the same conditions. Plots show cell
fractions gated on human CD45⫹ cells.
MEDICAL SCIENCES
colonies; when ⱖ5 secondary colonies were formed, 65% of
single IL-3R␣loCD45RA⫺ cells gave rise to both types of
colonies. Therefore, at least 28% (44 of 156) single IL3R␣loCD45RA⫺ cells that give rise to colonies in this assay have
bipotent granulocyte兾macrophage and megakar yocyte兾
erythrocyte potential. As with the cells plated from primary
short-term stromal cell culture, no CFU-GEMMs were obser ved, in contrast to 1.4% CFU-GEMMs from IL3R␣loCD45RA⫺ cells directly plated in methylcellulose, indicating the loss of multipotent differentiation capacity of single
cells in the primary culture. Based on these data, IL3R␣loCD45RA⫺ cells represent the CMP population.
CMPs, GMPs, and MEPs Contain No B or NK Cell-Restricted Progenitors.
To test B and NK cell developmental capacity of the three progenitor populations, cells were cultured for 3–5 weeks on Sys-1
stromal layers with SCF, FL, IL-7, and IL-2. Under these conditions
100 lymphoid lin⫺CD34⫹CD38⫹CD10⫹IL-7R␣⫹ progenitors reliably gave rise to B and NK cells, but from up to 5,000 CMPs, MEPs,
or GMPs no B or NK cell readout could be detected (Fig. 4).
However, up to 100 lin⫺ CD34⫹CD38⫺ cells tested did not give rise
to B or NK cells either, indicating that only lymphoid-committed
cells were able to readout into those lineages under the given culture
conditions. Therefore, although we cannot exclude a possibility that
the myeloid progenitors harbor cells with some lymphoid developmental potential in vivo, none of the three populations contains cells
readily committed to the B or NK lineage.
Myeloid Progenitors Have Limited or No Self-Renewing Capacity. To
test limited self-renewal or self-renewal activity, HSC-enriched cells
(lin⫺CD34⫹CD38⫺), CMPs, GMPs, and MEPs were plated in
limiting dilution (10–5,000 cells) in LTC-IC assays and transferred
to methylcellulose after 6 weeks of culture to evaluate myeloid
colony formation. Three experiments with different donors were
performed. The estimated frequency of LTC-IC was 1 in 20 for
lin⫺CD34⫹CD38⫺ cells, 1 in 1,700 for CMPs, 1 in 4,600 for GMPs,
and 1 in 5,800 for MEPs (data not shown). Therefore, myeloid
progenitor cells have only limited or no self-renewing capacity in
this assay.
Progenitor Populations Show Distinct, Lineage-Specific Gene Expression Profiles. To evaluate whether distinct progenitors express
specific transcription factors, cytokine receptors, enzymes, and
other functional proteins, we tested the expression of several
genes in HSC-enriched lin⫺CD34⫹CD38⫺ cells, myeloid progenitors, lin⫺CD34⫹CD38⫹CD10⫹ lymphoid progenitors, and
whole CD34⫹ cells by reverse transcription–PCR (Fig. 5). The
results show expression profiles generally consistent with the
distinct in vitro readout capacities of the different progenitors.
Manz et al.
Fig. 5. Differential mRNA expression of hematopoiesis-affiliated genes in
primitive progenitors (CD34⫹CD38⫺), myeloid-committed progenitors, and
lymphoid-committed progenitors. Total CD34⫹ bone marrow cells were used
as controls. GM-CSFR, granulocyte兾macrophage colony-stimulating factor
receptor; G-CSFR, granulocyte colony-stimulating factor receptor; GAPDH,
glyceraldehyde-3-phosphate dehydrogenase; vWG, von Willebrand factor;
C兾EBP␧, CCAAT-enhancer-binding protein ␧; MPO, myeloperoxidase; EpoR,
erythropoietin receptor; TdT, terminal deoxynucleotidyl transferase.
Whereas granulocyte colony-stimulating factor receptor,
C兾EBP␧, and MPO were expressed by GMPs and not by MEPs,
GATA-1, EpoR, c-mpl, ␤-globin, and von Willebrand factor
were detected in MEPs but not in GMPs. Interestingly, some
gene transcripts such as PU.1, MPO, and GATA-1 were detectable in HSC-enriched CD34⫹CD38⫺ cells. It shall be important
to test whether these transcripts result in protein production that
is involved in cell fate decision at this early developmental stage.
It shall also be important to determine whether these transcripts
are present in lin⫺CD34⫹CD38⫺Thy1⫹ HSC or in Thy1⫺ progenitors (17). Of note, none of the myeloid progenitors expressed
detectable levels of genes relevant in lymphoid development as
TdT, GATA-3, preT␣, IL-7R␣, and Pax-5, which were expressed in
the lymphoid-committed lin⫺CD34⫹CD38⫹CD10⫹ progenitors.
Therefore, the gene-expression profile supports the exclusive separation of the lymphoid and myeloid and further the downstream
granulocyte兾macrophage and megakaryocyte兾erythrocyte developmental pathway.
CMPs, GMPs, and MEPs Can Be Isolated from Cord Blood. Two cord
blood samples were evaluated for the above surface-marker
expression, and 100 cells of each population were tested for
PNAS 兩 September 3, 2002 兩 vol. 99 兩 no. 18 兩 11875
myeloid colony-formation activity. Although the distinct surface-marker expression profile was similar to adult bone marrow,
percentages of the myeloid progenitor populations were slightly
different: IL-3R␣loCD45RA⫺ CMPs account for ⬇0.4%, IL3R␣loCD45RA⫹ GMPs for ⬇0.3%, and IL-3R␣⫺CD45RA⫹
MEPs for ⬇0.05% of the mononuclear cell fraction (data not
shown). HSC-enriched lin⫺CD34⫹CD38⫺ cells and CMPs
formed all types of colonies with cloning efficiencies of 68 and
83%, respectively. GMPs formed exclusively granulocyte兾
macrophage colonies (cloning efficiency 41%), and MEPs
formed megakaryocyte兾erythrocyte colonies (cloning efficiency
88%) with only 4% granulocyte兾macrophage colony readout
(data not shown). Therefore, myeloid progenitor cells with
similar lineage restrictions can be found in cord blood.
Discussion
In this study we dissect by prospective isolation human multipotent hematopoietic progenitor cells from the earliest committed CMPs and further the downstream GMPs from MEPcommitted stages. We directly show that CMPs are clonogenic
progenitors of both GMPs and MEPs. The phenotype of these
populations does not overlap with that of HSCs, and the defined
populations have only very low LTC-IC activity. Furthermore,
none of these myeloid progenitors exhibit in vitro differentiation
activity for B or NK cells. We did not test the T cell potential of
these populations; however, myeloid progenitors express neither
early lymphoid markers such as CD7 and CD10 (7, 8) nor IL-7R␣
(not shown), which is expressed in human T cell progenitors
and transduces indispensable signals in human T cell development (25). In two experiments we transplanted 1.5 ⫻ 105
CD34⫹CD38⫺ HSC-enriched cells, 3.8 ⫻ 105 CMPs, and 1.5 ⫻
105 MEPs into sublethally irradiated NOD兾SCID兾␤2m⫺兾⫺ recipients. Three weeks after transplantation, mouse bone marrow
contained 21, 30, and 3% human CD45⫹ and兾or CD71⫹ nucleated cells derived from CD34⫹CD38⫺ cells, CMPs, and MEPs,
respectively. CD34⫹CD38⫺ cells and CMPs gave rise to both
glycophorin A⫹ erythroid cells and CD15⫹ myelomonocytic
cells, whereas MEPs exclusively gave rise to glycophorin A⫹
erythroid cells (not shown). Thus, considering their high cloning
efficiency, expansion capacity, and restricted differentiation
potentials, these populations are likely the human counterparts
of mouse myeloid progenitors. Although B cell potentials are
undetectable in myeloid progenitors in vitro, in vivo progeny of
3.8 ⫻ 105 CMPs contained a small number (0.1%) of CD19⫹ cells
(not shown). It is unclear whether the minor B cell progeny is
derived from contaminants of lymphoid precursors among a high
number of CMPs transplanted or whether putative bipotent B
cell兾myeloid progenitors share their surface phenotype with
CMPs. It shall be important to clarify this issue by future studies,
because minor B cell potentials were observed similarly in both
fetal and adult murine CMPs (2, 26).
The defined myeloid progenitors exhibit distinct surfaceexpression patterns of IL-3R␣ and CD45RA. IL-3 is an early
cytokine that supports proliferation and differentiation of HSCs
as well as primitive multipotent and myeloid progenitors (27–
29), whereas the receptor tyrosine phosphatase CD45, and
presumably its CD45RA isoform, can negatively regulate at least
some classes of cytokine-receptor signaling (30). Several previous studies have been communicated concerning the surfaceantigen expression of early myeloid progenitor cells in humans
(3, 6, 9–12). It was shown that CD34⫹CD45RA⫹ cells are devoid
of erythroid and enriched for granulocyte and macrophage
progenitor cells (6, 10). Also, different levels of IL-3R␣ expression on CD34⫹ cells reportedly are associated with preferential
lineage readout, as IL-3R␣⫺ cells are enriched for erythroid,
IL-3R␣lo cells for multipotent, and IL-3R␣⫹ cells for granulocyte兾macrophage colony-forming cells (9). However, in those
studies early lymphoid cells, lineage low or positive cells, and兾or
11876 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.172384399
HSCs and multipotent progenitors within the CD34⫹ fraction
were not excluded by using negative or positive markers, likely
resulting in heterogeneous populations. When reported, the
populations exhibited relatively low cloning efficiencies (6, 12) or
contained multipotent progenitors and HSCs (9). In our study,
the relationship between the expression pattern of CD45RA兾
IL-3R␣ and lineage potentials within the myeloid progenitor
fraction is consistent with previous reports. The additional use
of a panel of markers enabled us to highly purify lineal populations of CMPs and downstream GMPs and MEPs, which
represent developmental checkpoints for erythroid兾megakaryocyte versus granulocyte兾macrophage lineages.
It was reported also that CD64 (Fc␥R I)-expressing cells in the
human CD34⫹ bone marrow fraction (⬇23% of fetal bone marrow
CD34⫹ cells) contained granulocyte兾macrophage-committed progenitors at a relatively low frequency (cloning efficiency 26–50%;
refs. 11 and 12). We could not find CD64 expression on the
lin⫺CD34⫹CD38⫹ cells by using antibodies from hybridoma clone
10.1, which also were used in these previous studies (11). The most
likely explanation for this finding is that the CD64⫹ cells were gated
out in our study by using CD11b as a lineage marker, because the
vast majority of CD64⫹ cells reportedly coexpress CD11b (11).
Therefore it is possible that the CD64⫹ progenitors may reside in
a lineage low to positive fraction, which is likely to be downstream
of GMPs. Indeed, although the progenitor populations described
here cover most of the colony-forming cells, we found a low
frequency of both erythroid and myeloid colony-forming cells in the
CD34⫹CD38⫹lin⫹ fraction. We propose that most if not all myeloid
hematopoiesis is developing through the here-described progenitors; however, we cannot exclude the existence of other, paralleldeveloping, minor progenitor populations that express lineage
markers.
In mouse hematopoiesis, CMPs, GMPs and MEPs all are
contained in the lin⫺c-Kit⫹Sca-1⫺ population and are distinguished further based on the expression profile of CD34 and the
Fc␥R II兾III (CD16兾CD32; ref. 2). In contrast, all the human
myeloid progenitor populations described here are positive for
CD34 and do not express detectable levels of CD16兾CD32. We
therefore in turn looked for IL-3R␣ and CD45RA expression on
murine myeloid progenitors (2). Interestingly, in mouse hematopoiesis, IL-3R␣ as well as CD45RA expression both were
negative to low in MEPs, intermediate in CMPs, and positive in
GMPs, without providing sufficient discrimination to clearly set
a cut-off level between the different cellular fractions (not
shown). Thus, a number of markers expressed in progenitor or
stem cells seem to be differentially regulated in early human and
mouse myelopoiesis (31).
The prospectively purified, lineage-restricted human progenitors
at defined myeloid developmental stages shown here should provide tools to study transcriptional regulation as well as the role of
intrinsic and兾or extrinsic signals in lineage commitment (32). Also,
common developmental pathways or lineage origins of rare hematopoietic cells such as dendritic cells can be evaluated directly (33,
34). Finally, these progenitors might be useful in determining stages
at which leukemogenic events occur (17, 35–38).
We thank R. Negrin and the Stanford Bone Marrow Transplant Laboratory,
C. Ricordi and the Miami School of Medicine Hematopoietic Cell Processing Facility for bone marrow, and the Stanford Department of Obstetrics for cord blood supply. We thank M. Kondo for helpful discussions, L.
Jerabek for excellent laboratory management, S. Smith for antibody conjugation, L. Hidalgo and D. Escoto for animal care, and the Stanford
fluorescence-activated cell sorter facility for flow-cytometer maintenance.
This work was supported in part by a fellowship from the Deutsche
Krebshilfe (to M.G.M.), grants from the Claudia Adams Barr Program, the
Jose Carreras International Leukemia Foundation, the Damon Runyon
Cancer Research Foundation (to K.A.), and the Leukemia and Lymphoma
Society de Villiers International Achievement Award & Grant, and National Institutes of Health Leukemia Grant CA86017 (to I.L.W.).
Manz et al.
Manz et al.
21. Berardi, A. C., Meffre, E., Pflumio, F., Katz, A., Vainchenker, W., Schiff, C.
& Coulombel, L. (1997) Blood 89, 3554–3564.
22. Miller, J. S., McCullar, V., Punzel, M., Lemischka, I. R. & Moore, K. A. (1999)
Blood 93, 96–106.
23. Hao, Q. L., Smogorzewska, E. M., Barsky, L. W. & Crooks, G. M. (1998) Blood
91, 4145–4151.
24. Terstappen, L. W., Huang, S., Safford, M., Lansdorp, P. M. & Loken, M. R.
(1991) Blood 77, 1218–1227.
25. Puel, A., Ziegler, S. F., Buckley, R. H. & Leonard, W. J. (1998) Nat. Genet. 20,
394–397.
26. Traver, D., Miyamoto, T., Christensen, J., Iwasaki-Arai, J., Akashi, K. &
Weissman, I. L. (2001) Blood 98, 627–635.
27. Ogawa, M. (1993) Blood 81, 2844–2853.
28. Kimura, T., Sakabe, H., Tanimukai, S., Abe, T., Urata, Y., Yasukawa, K.,
Okano, A., Taga, T., Sugiyama, H., Kishimoto, T. & Sonoda, Y. (1997) Blood
90, 4767–4778.
29. Domen, J. & Weissman, I. L. (2000) J. Exp. Med. 192, 1707–1718.
30. Irie-Sasaki, J., Sasaki, T., Matsumoto, W., Opavsky, A., Cheng, M., Welstead,
G., Griffiths, E., Krawczyk, C., Richardson, C. D., Aitken, K., et al. (2001)
Nature (London) 409, 349–354.
31. Okuno, Y., Iwasaki, H., Huettner, C. S., Radomska, H. S., Gonzalez, D. A.,
Tenen, D. G. & Akashi, K. (2002) Proc. Natl. Acad. Sci. USA 99, 6246–6251.
32. Kondo, M., Scherer, D. C., Miyamoto, T., King, A. G., Akashi, K., Sugamura,
K. & Weissman, I. L. (2000) Nature (London) 407, 383–386.
33. Traver, D., Akashi, K., Manz, M., Merad, M., Miyamoto, T., Engleman, E. G.
& Weissman, I. L. (2000) Science 290, 2152–2154.
34. Manz, M. G., Traver, D., Miyamoto, T., Weissman, I. L. & Akashi, K. (2001)
Blood 97, 3333–3341.
35. Reya, T., Morrison, S. J., Clarke, M. F. & Weissman, I. L. (2001) Nature
(London) 414, 105–111.
36. Traver, D., Akashi, K., Weissman, I. L. & Lagasse, E. (1998) Immunity 9, 47–57.
37. Kogan, S. C., Brown, D. E., Shultz, D. B., Truong, B. T., LallemandBreitenbach, V., Guillemin, M. C., Lagasse, E., Weissman, I. L. & Bishop, J. M.
(2001) J. Exp. Med. 193, 531–543.
38. Yuan, Y., Zhou, L., Miyamoto, T., Iwasaki, H., Harakawa, N., Hetherington,
C. J., Burel, S. A., Lagasse, E., Weissman, I. L., Akashi, K. & Zhang, D. E.
(2001) Proc. Natl. Acad. Sci. USA 98, 10398–10403.
PNAS 兩 September 3, 2002 兩 vol. 99 兩 no. 18 兩 11877
MEDICAL SCIENCES
1. Kondo, M., Weissman, I. L. & Akashi, K. (1997) Cell 91, 661–672.
2. Akashi, K., Traver, D., Miyamoto, T. & Weissman, I. L. (2000) Nature (London)
404, 193–197.
3. de Wynter, E. A., Heyworth, C. M., Mukaida, N., Jaworska, E., Weffort-Santos,
A., Matushima, K. & Testa, N. G. (2001) J. Leukocyte Biol. 70, 455–460.
4. Debili, N., Coulombel, L., Croisille, L., Katz, A., Guichard, J., Breton-Gorius,
J. & Vainchenker, W. (1996) Blood 88, 1284–1296.
5. Debili, N., Robin, C., Schiavon, V., Letestu, R., Pflumio, F., Mitjavila-Garcia,
M. T., Coulombel, L. & Vainchenker, W. (2001) Blood 97, 2023–2030.
6. Fritsch, G., Buchinger, P., Printz, D., Fink, F. M., Mann, G., Peters, C., Wagner,
T., Adler, A. & Gadner, H. (1993) Blood 81, 2301–2309.
7. Galy, A., Travis, M., Cen, D. & Chen, B. (1995) Immunity 3, 459–473.
8. Hao, Q. L., Zhu, J., Price, M. A., Payne, K. J., Barsky, L. W. & Crooks, G. M.
(2001) Blood 97, 3683–3690.
9. Huang, S., Chen, Z., Yu, J. F., Young, D., Bashey, A., Ho, A. D. & Law, P.
(1999) Stem Cells 17, 265–272.
10. Lansdorp, P. M., Sutherland, H. J. & Eaves, C. J. (1990) J. Exp. Med. 172,
363–366.
11. Olweus, J., Lund-Johansen, F. & Terstappen, L. W. (1995) Blood 85, 2402–
2413.
12. Olweus, J., Thompson, P. A. & Lund-Johansen, F. (1996) Blood 88, 3741–3754.
13. Roy, V., Miller, J. S. & Verfaillie, C. M. (1997) Exp. Hematol. (Charlottesville,
Va) 25, 387–394.
14. Ryan, D. H., Nuccie, B. L., Ritterman, I., Liesveld, J. L., Abboud, C. N. & Insel,
R. A. (1997) Blood 89, 929–940.
15. Clay, D., Rubinstein, E., Mishal, Z., Anjo, A., Prenant, M., Jasmin, C.,
Boucheix, C. & Le Bousse-Kerdiles, M. C. (2001) Blood 97, 1982–1989.
16. Rappold, I., Ziegler, B. L., Kohler, I., Marchetto, S., Rosnet, O., Birnbaum, D.,
Simmons, P. J., Zannettino, A. C., Hill, B., Neu, S., et al. (1997) Blood 90,
111–125.
17. Miyamoto, T., Weissman, I. L. & Akashi, K. (2000) Proc. Natl. Acad. Sci. USA
97, 7521–7526.
18. Petzer, A. L., Hogge, D. E., Landsdorp, P. M., Reid, D. S. & Eaves, C. J. (1996)
Proc. Natl. Acad. Sci. USA 93, 1470–1474.
19. Bhatia, M., Wang, J. C., Kapp, U., Bonnet, D. & Dick, J. E. (1997) Proc. Natl.
Acad. Sci. USA 94, 5320–5325.
20. Baum, C. M., Weissman, I. L., Tsukamoto, A. S., Buckle, A. M. & Peault, B.
(1992) Proc. Natl. Acad. Sci. USA 89, 2804–2808.